Cyberbullying Prediction As Cyber Counseling Tools With Data Mining Classification

Abstract

The growth of data and information is increasing rapidly while the users of information technology devices continue to increase. Moreover, the ease of access to information is supported by the presence of mobile communication technology which is owned by almost every user. Currently, there is a significant increase in the number of users, who are considered to have the opportunity for the presence of cybercrimes, especially in cases of bullying. The main problem is that it is difficult to predict the potential for cyberbullying because cyberspace is full of anonymity. Cases of cyberbullying include cyber crimes outside of computer security when viewed from a behavioral perspective. Therefore, cyberbullying analyzes behavior when carrying out negative actions. In this study, we have investigated and predicted bullying tendencies using a data mining approach. Analysis of the presented case studies and their technical implementation through a data mining approach. Using data mining as a tool, we use several classification methods with the aim of comparison and which method is best for analysis. Classification methods in data mining are K-NN, Random Forest, Decision Tree, and Naive Bayes. By comparison of the four methods, there are three classes. There are three classes, namely no potential for bullying, violence and insults. Thus, three classes were detected based on the results of the investigations and several techniques were evaluated using a comparison of the performance of each technique. The final result will show the best performing decision tree technique because during the preprocessing stage it does not take long. the addition is because of the ability to break down each branch of the data so that the time required for data cleaning is faster with an acceleration of 7%.